<ul><li>Researchers have introduced a rehearsal-free federated domain-incremental learning framework called RefFiL.</li><li>RefFiL is designed to address the challenge of catastrophic forgetting in federated domain-incremental learning.</li><li>The framework learns domain-invariant knowledge and incorporates domain-specific prompts from different federated learning participants.</li><li>RefFiL effectively mitigates forgetting without requiring extra memory space, making it suitable for privacy-sensitive and resource-constrained devices.</li></ul>

Rehearsal-free Federated Domain-incremental Learning

Discover more